1
|
/**
|
2
|
* '$RCSfile$'
|
3
|
* Purpose: A class to asyncronously do delta-T replication checking
|
4
|
* Copyright: 2000 Regents of the University of California and the
|
5
|
* National Center for Ecological Analysis and Synthesis
|
6
|
* Authors: Chad Berkley
|
7
|
*
|
8
|
* '$Author: leinfelder $'
|
9
|
* '$Date: 2011-11-02 20:40:12 -0700 (Wed, 02 Nov 2011) $'
|
10
|
* '$Revision: 6595 $'
|
11
|
*
|
12
|
* This program is free software; you can redistribute it and/or modify
|
13
|
* it under the terms of the GNU General Public License as published by
|
14
|
* the Free Software Foundation; either version 2 of the License, or
|
15
|
* (at your option) any later version.
|
16
|
*
|
17
|
* This program is distributed in the hope that it will be useful,
|
18
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
19
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
20
|
* GNU General Public License for more details.
|
21
|
*
|
22
|
* You should have received a copy of the GNU General Public License
|
23
|
* along with this program; if not, write to the Free Software
|
24
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
25
|
*/
|
26
|
|
27
|
package edu.ucsb.nceas.metacat;
|
28
|
|
29
|
import java.io.StringReader;
|
30
|
import java.sql.PreparedStatement;
|
31
|
import java.sql.ResultSet;
|
32
|
import java.sql.SQLException;
|
33
|
import java.util.Hashtable;
|
34
|
import java.util.Vector;
|
35
|
|
36
|
import org.apache.log4j.Logger;
|
37
|
import org.apache.xpath.objects.XObject;
|
38
|
import org.apache.xpath.XPathAPI;
|
39
|
import org.apache.xerces.parsers.DOMParser;
|
40
|
import org.apache.xerces.dom.DocumentTypeImpl;
|
41
|
import org.w3c.dom.Attr;
|
42
|
import org.w3c.dom.NamedNodeMap;
|
43
|
import org.w3c.dom.NodeList;
|
44
|
import org.w3c.dom.Document;
|
45
|
import org.w3c.dom.Node;
|
46
|
import org.w3c.dom.NodeList;
|
47
|
import org.w3c.dom.DocumentType;
|
48
|
import org.xml.sax.InputSource;
|
49
|
|
50
|
import javax.xml.parsers.DocumentBuilder;
|
51
|
import javax.xml.parsers.DocumentBuilderFactory;
|
52
|
import javax.xml.parsers.ParserConfigurationException;
|
53
|
import javax.xml.transform.*;
|
54
|
import javax.xml.transform.stream.*;
|
55
|
import javax.xml.transform.dom.*;
|
56
|
|
57
|
import org.ecoinformatics.eml.EMLParser;
|
58
|
|
59
|
import edu.ucsb.nceas.metacat.database.DBConnection;
|
60
|
import edu.ucsb.nceas.metacat.database.DBConnectionPool;
|
61
|
import edu.ucsb.nceas.metacat.properties.PropertyService;
|
62
|
import edu.ucsb.nceas.metacat.util.DocumentUtil;
|
63
|
import edu.ucsb.nceas.metacat.util.MetacatUtil;
|
64
|
import edu.ucsb.nceas.utilities.PropertyNotFoundException;
|
65
|
/**
|
66
|
* This class will figure out which content type it is for a given data file.
|
67
|
* First, from xml_relation to get all relative files to this data file.
|
68
|
* Then from xml_documents to get physical files. From physical file pull out
|
69
|
* the content type
|
70
|
*/
|
71
|
public class ContentTypeProvider
|
72
|
{
|
73
|
private String dataFileId = null;
|
74
|
private String contentType = null;
|
75
|
private String packageType = null;
|
76
|
private Hashtable contentTypeHash = new Hashtable();
|
77
|
|
78
|
//Constant
|
79
|
private String BETA = "beta";
|
80
|
private String EML2 = "eml2";
|
81
|
private static String DEFAULTCONTENTTYPE;
|
82
|
static {
|
83
|
try {
|
84
|
DEFAULTCONTENTTYPE = PropertyService.getProperty("replication.defaultcontenttype");
|
85
|
} catch (PropertyNotFoundException pnfe) {
|
86
|
System.err.println("Could not get property DEFAULTCONTENTTYPE:"
|
87
|
+ pnfe.getMessage());
|
88
|
}
|
89
|
}
|
90
|
private String FORMATPATH = "//format";
|
91
|
private String TEXT = "text";
|
92
|
private String TEXTYPE ="text/plain";
|
93
|
private String XML = "xml";
|
94
|
private String XMLTYPE = "text/xml";
|
95
|
private String HTML = "HTML";
|
96
|
private String HTMLTYPE = "text/html";
|
97
|
private String GIF = "gif";
|
98
|
private String JPEG = "jpeg";
|
99
|
private String JPEGTYPE = "image/jpeg";
|
100
|
private String GIFTYPE = "image/gif";
|
101
|
private String BMP = "bmp";
|
102
|
private String BMPTYPE = "image/bmp";
|
103
|
private String TAR = "tar";
|
104
|
private String TARTYPE ="application/x-tar";
|
105
|
private String ZIP = "zip";
|
106
|
private String ZIPTYPE = "application/x-zip-compressed";
|
107
|
private String BINARY = "binary";
|
108
|
private String BINARYTYPE = "application/octet-stream";
|
109
|
|
110
|
private String ENTITYDOCTYPE = "xml.entitydoctype";
|
111
|
private String PHYSICALDOCTYPE = "xml.physicaldoctype";
|
112
|
private String EML2DOCTYPE = "eml2namespace";
|
113
|
private String DATAFORMAT = "dataFormat";
|
114
|
private String TEXTFORMAT = "textFormat";
|
115
|
private String EXTENALFORMAT = "externallyDefinedFormat";
|
116
|
private String FORMATNAME = "formatName";
|
117
|
private String BINARYRASTERFORMAT = "binaryRasterFormat";
|
118
|
|
119
|
private String DATAFILEPATH ="//physical/distribution/online/url";
|
120
|
private static Logger logMetacat = Logger.getLogger(ContentTypeProvider.class);
|
121
|
|
122
|
/**
|
123
|
* Constructor of ContentTypeProvider
|
124
|
*/
|
125
|
public ContentTypeProvider(String docIdWithRevision)
|
126
|
{
|
127
|
dataFileId = DocumentUtil.getDocIdFromString(docIdWithRevision);
|
128
|
//get relative doclist for data file and package type
|
129
|
Vector docLists = null;
|
130
|
docLists = getRelativeDocIdList(dataFileId);
|
131
|
|
132
|
if ( packageType == null)
|
133
|
{
|
134
|
// other situation, contenetype is default value
|
135
|
contentType = DEFAULTCONTENTTYPE;
|
136
|
}
|
137
|
else if (packageType.equals(BETA))
|
138
|
{
|
139
|
// for beta package and get entity docid for the data file
|
140
|
String entityDocid = getTargetDocIdForBeta(docLists, ENTITYDOCTYPE);
|
141
|
// get physical docid for data file
|
142
|
docLists = getRelativeDocIdList(entityDocid);
|
143
|
String physicalDocId = getTargetDocIdForBeta(docLists, PHYSICALDOCTYPE);
|
144
|
// if no physical docid assign to this data file, content type is default
|
145
|
if (physicalDocId == null)
|
146
|
{
|
147
|
|
148
|
contentType = DEFAULTCONTENTTYPE;
|
149
|
}
|
150
|
else
|
151
|
{
|
152
|
|
153
|
parsePhysicalDocumentForBeta(physicalDocId);
|
154
|
}
|
155
|
}
|
156
|
else if (packageType.equals(EML2))
|
157
|
{
|
158
|
// for eml2 package
|
159
|
// get eml document for data file
|
160
|
//String eml2Docid = getTargetDocIdForBeta(docLists, EML2DOCTYPE);
|
161
|
String eml2Docid = (String)docLists.elementAt(0);
|
162
|
findContentTypeInEML2(eml2Docid);
|
163
|
|
164
|
}
|
165
|
|
166
|
}
|
167
|
|
168
|
/** Method to get content type */
|
169
|
public String getContentType()
|
170
|
{
|
171
|
return contentType;
|
172
|
}//getContentType
|
173
|
|
174
|
/* Method to find content type base on data format*/
|
175
|
private void findContentTypeInEML2(String eml2DocId)
|
176
|
{
|
177
|
if (eml2DocId == null)
|
178
|
{
|
179
|
contentType = DEFAULTCONTENTTYPE;
|
180
|
return;
|
181
|
}
|
182
|
DocumentImpl xmlDoc = null;
|
183
|
String xmlString = null;
|
184
|
StringReader read = null;
|
185
|
InputSource in = null;
|
186
|
DocumentBuilderFactory dfactory = null;
|
187
|
Document doc = null;
|
188
|
// create xml document
|
189
|
try
|
190
|
{
|
191
|
String accNumber = eml2DocId + PropertyService.getProperty("document.accNumSeparator") +
|
192
|
DBUtil.getLatestRevisionInDocumentTable(eml2DocId);
|
193
|
//System.out.println("the acc number is !!!!!!!!!!!!!!!!!"+accNumber);
|
194
|
xmlDoc = new DocumentImpl(accNumber);
|
195
|
xmlString = xmlDoc.toString();
|
196
|
//System.out.println("the xml doc is "+xmlDoc);
|
197
|
// create dom tree
|
198
|
read = new StringReader(xmlString);
|
199
|
in = new InputSource(read);
|
200
|
dfactory = DocumentBuilderFactory.newInstance();
|
201
|
dfactory.setNamespaceAware(false);
|
202
|
doc = dfactory.newDocumentBuilder().parse(in);
|
203
|
}
|
204
|
catch (Exception e)
|
205
|
{
|
206
|
// if faild, set default value
|
207
|
contentType = DEFAULTCONTENTTYPE;
|
208
|
logMetacat.error("Error in ContentTypeProvider." +
|
209
|
"findContentTypeInEML2()" + e.getMessage());
|
210
|
return;
|
211
|
}
|
212
|
Node dataFormatNode = findDataFormatNodeInEML2(doc, DATAFILEPATH,
|
213
|
dataFileId);
|
214
|
if (dataFormatNode == null)
|
215
|
{
|
216
|
contentType = DEFAULTCONTENTTYPE;
|
217
|
logMetacat.info("Couldn't find data format node");
|
218
|
return;
|
219
|
|
220
|
}
|
221
|
NodeList childList = dataFormatNode.getChildNodes();
|
222
|
// go through childList
|
223
|
for (int i = 0; i<childList.getLength(); i++)
|
224
|
{
|
225
|
Node child = childList.item(i);
|
226
|
|
227
|
// if has text format child set to text/plain
|
228
|
if (child.getNodeName() != null && child.getNodeName().equals(TEXTFORMAT))
|
229
|
{
|
230
|
logMetacat.info("in text format");
|
231
|
contentType = TEXTYPE;
|
232
|
}
|
233
|
|
234
|
//external format
|
235
|
if (child.getNodeName() != null && child.getNodeName().equals(EXTENALFORMAT))
|
236
|
{
|
237
|
logMetacat.info("in external format ");
|
238
|
String format = getTextValueForGivenChildTag(child, FORMATNAME);
|
239
|
logMetacat.info("The format is: "+format);
|
240
|
// if we can find the format in the contentTypeHash table
|
241
|
contentType = (String)lookUpContentType(format);
|
242
|
if (contentType == null)
|
243
|
{
|
244
|
contentType = BINARYTYPE;
|
245
|
}
|
246
|
}
|
247
|
|
248
|
// binaryRasterFormat
|
249
|
if (child.getNodeName() != null && child.getNodeName().
|
250
|
equals(BINARYRASTERFORMAT))
|
251
|
{
|
252
|
contentType = BINARYTYPE;
|
253
|
}//if
|
254
|
}//for
|
255
|
//if contentype still be null, set default value
|
256
|
if (contentType == null)
|
257
|
{
|
258
|
contentType = DEFAULTCONTENTTYPE;
|
259
|
}
|
260
|
}
|
261
|
|
262
|
/* Method get text value of given child tagname*/
|
263
|
private String getTextValueForGivenChildTag(Node parentNode,
|
264
|
String childTagName)
|
265
|
{
|
266
|
String textValue = null;
|
267
|
NodeList childList = parentNode.getChildNodes();
|
268
|
for (int i= 0; i<childList.getLength();i++)
|
269
|
{
|
270
|
Node child = childList.item(i);
|
271
|
if (child.getNodeName() != null && child.getNodeName().equals(childTagName))
|
272
|
{
|
273
|
logMetacat.info("Find child node: " + childTagName);
|
274
|
Node textNode = child.getFirstChild();
|
275
|
if (textNode.getNodeType() == Node.TEXT_NODE)
|
276
|
{
|
277
|
textValue = textNode.getNodeValue();
|
278
|
}//if
|
279
|
}//if
|
280
|
}//for
|
281
|
logMetacat.info("The text value for element- " + childTagName +
|
282
|
" is " + textValue);
|
283
|
return textValue;
|
284
|
}//getTExtValueForGivenChildTag
|
285
|
|
286
|
/* Find the data format node in eml2 document */
|
287
|
private Node findDataFormatNodeInEML2(Document xml, String xPath,
|
288
|
String targetDocId)
|
289
|
{
|
290
|
Node targetNode = null;
|
291
|
Node node = findDataFileNodeInEML2(xml, xPath, targetDocId);
|
292
|
if (node != null)
|
293
|
{
|
294
|
// get the phycial the prent is online, grandparent is distribution
|
295
|
// the grand'parent is physical
|
296
|
Node phyicalNode = node.getParentNode().getParentNode().getParentNode();
|
297
|
NodeList list = phyicalNode.getChildNodes();
|
298
|
for (int i = 0; i < list.getLength(); i++)
|
299
|
{
|
300
|
Node kid = list.item(i);
|
301
|
// find dataFormat node
|
302
|
if (kid.getNodeType() == node.ELEMENT_NODE &&
|
303
|
kid.getNodeName().equals(DATAFORMAT))
|
304
|
{
|
305
|
targetNode = kid;
|
306
|
break;
|
307
|
} //if
|
308
|
} //for
|
309
|
if (targetNode != null)
|
310
|
{
|
311
|
logMetacat.info("dataFormat node'name: " +
|
312
|
targetNode.getNodeName());
|
313
|
}
|
314
|
}//if
|
315
|
return targetNode;
|
316
|
}
|
317
|
/* Find the datafile node */
|
318
|
private Node findDataFileNodeInEML2(Document xml, String xPath,
|
319
|
String targetDocId)
|
320
|
{
|
321
|
Node dataFileNode = null;
|
322
|
NodeList list = null;
|
323
|
try
|
324
|
{
|
325
|
list = XPathAPI.selectNodeList(xml, xPath);
|
326
|
}
|
327
|
catch (Exception e)
|
328
|
{
|
329
|
// catch an error and return null
|
330
|
logMetacat.error("Error in findDataFileNode: "+e.getMessage());
|
331
|
return dataFileNode;
|
332
|
}
|
333
|
// go through the list and find target docid in online/url
|
334
|
if (list != null)
|
335
|
{
|
336
|
for (int i = 0; i < list.getLength(); i++)
|
337
|
{
|
338
|
Node node = list.item(i);
|
339
|
Node textNode = node.getFirstChild();
|
340
|
if (textNode.getNodeType() == node.TEXT_NODE)
|
341
|
{
|
342
|
String URLData = textNode.getNodeValue();
|
343
|
logMetacat.info("online/url text data: " + URLData);
|
344
|
//Only handle ecogrid data file
|
345
|
if (URLData.indexOf(DBSAXHandler.ECOGRID) != -1 )
|
346
|
{
|
347
|
// Get docid from url
|
348
|
String docId =
|
349
|
DocumentUtil.getAccessionNumberFromEcogridIdentifier(URLData);
|
350
|
// Get rid of revision
|
351
|
docId = DocumentUtil.getDocIdFromAccessionNumber(docId);
|
352
|
logMetacat.info("docid from url element in xml is: " +
|
353
|
docId);
|
354
|
//if this docid equals target one, we find it
|
355
|
if (docId != null && docId.equals(targetDocId))
|
356
|
{
|
357
|
logMetacat.info("Find target docid in online/url: " +
|
358
|
docId);
|
359
|
dataFileNode = node;
|
360
|
break;
|
361
|
}
|
362
|
} //if
|
363
|
|
364
|
} //if
|
365
|
} //for
|
366
|
}//if
|
367
|
|
368
|
return dataFileNode;
|
369
|
}//findDataFileNode
|
370
|
|
371
|
/* Get relative docid list and packagetype */
|
372
|
private Vector getRelativeDocIdList(String id)
|
373
|
{
|
374
|
Vector docList = new Vector();
|
375
|
String sql = "SELECT packagetype, subject from xml_relation " +
|
376
|
"where object = ?";
|
377
|
ResultSet rs = null;
|
378
|
PreparedStatement pStmt=null;
|
379
|
DBConnection conn = null;
|
380
|
int serialNumber = -1;
|
381
|
try
|
382
|
{
|
383
|
//check out DBConnection
|
384
|
conn=DBConnectionPool.getDBConnection
|
385
|
("ContentTypeProvider.getRelativeDocIdlist");
|
386
|
serialNumber=conn.getCheckOutSerialNumber();
|
387
|
pStmt = conn.prepareStatement(sql);
|
388
|
// binding value
|
389
|
pStmt.setString(1, id);
|
390
|
//execute query
|
391
|
pStmt.execute();
|
392
|
rs = pStmt.getResultSet();
|
393
|
// get result list
|
394
|
String packType = null;
|
395
|
while (rs.next())
|
396
|
{
|
397
|
packType = rs.getString(1);
|
398
|
String subject = rs.getString(2);
|
399
|
|
400
|
// get rid of duplicate record and add the docid into vector
|
401
|
if (!docList.contains(subject))
|
402
|
{
|
403
|
|
404
|
docList.add(subject);
|
405
|
}
|
406
|
}//while
|
407
|
|
408
|
// set up data package type
|
409
|
if ((MetacatUtil.getOptionList(PropertyService.getProperty("xml.packagedoctype"))).
|
410
|
contains(packType))
|
411
|
{
|
412
|
//this is beta4 or beta6 version
|
413
|
logMetacat.info("This is beta package");
|
414
|
packageType = BETA;
|
415
|
}
|
416
|
else if ((MetacatUtil.getOptionList
|
417
|
(PropertyService.getProperty("xml.eml2_0_0namespace"))).contains(packType))
|
418
|
{
|
419
|
// this eml 2 document
|
420
|
logMetacat.info("This is EML2.0.0 package");
|
421
|
packageType = EML2;
|
422
|
}
|
423
|
else if ((MetacatUtil.getOptionList
|
424
|
(PropertyService.getProperty("xml.eml2_0_1namespace"))).contains(packType))
|
425
|
{
|
426
|
// this eml 2 document
|
427
|
logMetacat.info("This is EML2.0.1 package");
|
428
|
packageType = EML2;
|
429
|
}
|
430
|
|
431
|
|
432
|
|
433
|
}//try
|
434
|
catch(SQLException e)
|
435
|
{
|
436
|
|
437
|
logMetacat.error("ContenTypProvider.getRelativeDoclist1 " +
|
438
|
e.getMessage());
|
439
|
}//catch
|
440
|
catch(PropertyNotFoundException pnfe)
|
441
|
{
|
442
|
logMetacat.error("ContenTypProvider.getRelativeDoclist1 " +
|
443
|
pnfe.getMessage());
|
444
|
}//catch
|
445
|
finally
|
446
|
{
|
447
|
try
|
448
|
{
|
449
|
pStmt.close();
|
450
|
}
|
451
|
catch (SQLException ee)
|
452
|
{
|
453
|
logMetacat.error("ContenTypProvider.getRelativeDoclist2 " +
|
454
|
ee.getMessage());
|
455
|
}
|
456
|
finally
|
457
|
{
|
458
|
DBConnectionPool.returnDBConnection(conn, serialNumber);
|
459
|
}
|
460
|
}//finally
|
461
|
|
462
|
return docList;
|
463
|
}// getRelativeDocIdList
|
464
|
|
465
|
/* Method to get physical document for data file in xml_documents table for
|
466
|
* beta eml package
|
467
|
*/
|
468
|
private String getTargetDocIdForBeta(Vector list, String targetType)
|
469
|
{
|
470
|
String docId = null;
|
471
|
// make sure list is not empty
|
472
|
if (list.isEmpty())
|
473
|
{
|
474
|
|
475
|
return docId;
|
476
|
}
|
477
|
// get sql command
|
478
|
String sql = "SELECT doctype, docid from xml_documents where docid in ( ";
|
479
|
// the first element
|
480
|
sql = sql + "?";
|
481
|
// remaining values
|
482
|
for (int i = 1; i < list.size(); i++) {
|
483
|
sql = sql + ", ?";
|
484
|
}
|
485
|
// add parentheses
|
486
|
sql = sql + ")";
|
487
|
logMetacat.info("SQL for select doctype: "+ sql);
|
488
|
ResultSet rs = null;
|
489
|
PreparedStatement pStmt=null;
|
490
|
DBConnection conn = null;
|
491
|
int serialNumber = -1;
|
492
|
try
|
493
|
{
|
494
|
//check out DBConnection
|
495
|
conn=DBConnectionPool.getDBConnection
|
496
|
("ContentTypeProvider.setPhycialDocIdForBeta");
|
497
|
serialNumber=conn.getCheckOutSerialNumber();
|
498
|
pStmt = conn.prepareStatement(sql);
|
499
|
// set the parameter values
|
500
|
for (int i = 0; i < list.size(); i++) {
|
501
|
String docid = (String) list.elementAt(i);
|
502
|
pStmt.setString(i+1, docid);
|
503
|
}
|
504
|
//execute query
|
505
|
pStmt.execute();
|
506
|
rs = pStmt.getResultSet();
|
507
|
// get result list
|
508
|
while (rs.next())
|
509
|
{
|
510
|
String packType = rs.getString(1);
|
511
|
String targetId = rs.getString(2);
|
512
|
// find physical document
|
513
|
if ((MetacatUtil.getOptionList(PropertyService.getProperty(targetType))).
|
514
|
contains(packType))
|
515
|
{
|
516
|
// assign physical document and jump out the while loop
|
517
|
docId = targetId;
|
518
|
break;
|
519
|
}
|
520
|
}//while
|
521
|
|
522
|
}//try
|
523
|
catch(SQLException e)
|
524
|
{
|
525
|
|
526
|
logMetacat.error("ContenTypProvider.setPhysicalDocIdForBeta1 " +
|
527
|
e.getMessage());
|
528
|
}//catch
|
529
|
catch(PropertyNotFoundException pnfe)
|
530
|
{
|
531
|
|
532
|
logMetacat.error("ContenTypProvider.setPhysicalDocIdForBeta1 " +
|
533
|
pnfe.getMessage());
|
534
|
}//catch
|
535
|
finally
|
536
|
{
|
537
|
try
|
538
|
{
|
539
|
pStmt.close();
|
540
|
}
|
541
|
catch(SQLException ee)
|
542
|
{
|
543
|
logMetacat.error("ContenTypProvider.setPhysicalDocIdForBeta2 " +
|
544
|
ee.getMessage());
|
545
|
}//catch
|
546
|
finally
|
547
|
{
|
548
|
DBConnectionPool.returnDBConnection(conn, serialNumber);
|
549
|
}
|
550
|
}//finally
|
551
|
logMetacat.info("target docid is: "+ docId + " "+
|
552
|
"for target doctype: "+targetType);
|
553
|
return docId;
|
554
|
}
|
555
|
|
556
|
|
557
|
|
558
|
|
559
|
/* Parser the beta physical document and find the value in format element*/
|
560
|
private void parsePhysicalDocumentForBeta(String physicalDocid)
|
561
|
{
|
562
|
String xmlDoc = null;
|
563
|
try
|
564
|
{
|
565
|
String accNumber = physicalDocid + PropertyService.getProperty("document.accNumSeparator") +
|
566
|
DBUtil.getLatestRevisionInDocumentTable(physicalDocid);
|
567
|
//System.out.println("the accenumber is !!!!!!!!!!!!!!!!!!" + accNumber);
|
568
|
DocumentImpl doc = new DocumentImpl(accNumber);
|
569
|
xmlDoc = doc.toString();
|
570
|
//System.out.println("The physical xml is "+xmlDoc);
|
571
|
}
|
572
|
catch (Exception e)
|
573
|
{
|
574
|
contentType = DEFAULTCONTENTTYPE;
|
575
|
logMetacat.error("Error in ContentTypeProvider." +
|
576
|
"parsePhysicalDocumentForBeta()" + e.getMessage());
|
577
|
return;
|
578
|
}
|
579
|
// get format element's text value
|
580
|
String format = getTextValueFromPath(new StringReader(xmlDoc), FORMATPATH);
|
581
|
|
582
|
if (format == null)
|
583
|
{
|
584
|
// if couldn't find the format, set contentype default value;
|
585
|
contentType = DEFAULTCONTENTTYPE;
|
586
|
}
|
587
|
else
|
588
|
{
|
589
|
// if can find a format and look up from hash to get value
|
590
|
contentType = lookUpContentType(format);
|
591
|
// couldn't find the content type for this format in hash table
|
592
|
if (contentType == null)
|
593
|
{
|
594
|
//set default vlaue
|
595
|
contentType = DEFAULTCONTENTTYPE;
|
596
|
}//if
|
597
|
}//else
|
598
|
}//parsePhysicalDocumentForBeta
|
599
|
|
600
|
private String getTextValueFromPath(StringReader xml, String xPath)
|
601
|
{
|
602
|
String textValue = null;
|
603
|
// get nodelist from doc by path
|
604
|
try
|
605
|
{
|
606
|
NodeList list = EMLParser.getPathContent(xml, xPath);
|
607
|
Node elementNode = list.item(0);
|
608
|
Node textNode = elementNode.getFirstChild();
|
609
|
if (textNode.getNodeType() == Node.TEXT_NODE)
|
610
|
{
|
611
|
textValue = textNode.getNodeValue();// get value
|
612
|
}
|
613
|
|
614
|
}
|
615
|
catch (Exception e)
|
616
|
{
|
617
|
logMetacat.error("error in ContentTypeProvider."+
|
618
|
"getTextValueFromPath: "+e.getMessage());
|
619
|
}
|
620
|
logMetacat.info("The text value for " + xPath + " is: "+
|
621
|
textValue);
|
622
|
return textValue;
|
623
|
}//getTextValueFromPath
|
624
|
|
625
|
/* A method to look up contentype */
|
626
|
private String lookUpContentType(String format)
|
627
|
{
|
628
|
String newFormat = null;
|
629
|
constructContentHashTable();
|
630
|
newFormat = format.toLowerCase().trim();
|
631
|
String type = null;
|
632
|
type = (String)contentTypeHash.get(newFormat);
|
633
|
logMetacat.info("contentType looked from hashtalbe is: " +
|
634
|
type);
|
635
|
return type;
|
636
|
}// lookupcontentypes
|
637
|
|
638
|
/* Construct content type hashtable */
|
639
|
private void constructContentHashTable()
|
640
|
{
|
641
|
contentTypeHash.put(TEXT, TEXTYPE);
|
642
|
contentTypeHash.put(XML, XMLTYPE);
|
643
|
contentTypeHash.put(HTML,HTMLTYPE);
|
644
|
contentTypeHash.put(GIF, GIFTYPE);
|
645
|
contentTypeHash.put(JPEG, JPEGTYPE);
|
646
|
contentTypeHash.put(BMP, BMPTYPE);
|
647
|
contentTypeHash.put(TAR, TARTYPE);
|
648
|
contentTypeHash.put(ZIP, ZIPTYPE);
|
649
|
contentTypeHash.put(BINARY, BINARYTYPE);
|
650
|
|
651
|
}//constructrContentHashTable();
|
652
|
|
653
|
|
654
|
|
655
|
public static void main(String[] argus)
|
656
|
{
|
657
|
try
|
658
|
{
|
659
|
DBConnectionPool pool = DBConnectionPool.getInstance();
|
660
|
//ContentTypeProvider provider = new ContentTypeProvider("tao.9830");
|
661
|
ContentTypeProvider provider = new ContentTypeProvider("tao.0001");
|
662
|
String str = provider.getContentType();
|
663
|
logMetacat.info("content type is : " + str);
|
664
|
}
|
665
|
catch(Exception e)
|
666
|
{
|
667
|
logMetacat.error("erorr in Schemalocation.main: " +
|
668
|
e.getMessage());
|
669
|
}
|
670
|
}
|
671
|
}//ContentTypeProvider
|